LinkedSpending: OpenSpending becomes Linked Open Data Konrad H¨ offner October 5, 2013 Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
Open Spending Data government spending data available to everyone what amount of money spent for which purpose, when and by which department demanded by the public increases accountability reduces corruption voters can make better informed decisions → strengthens democracy strengthens the government itself, more likely to commit to large projects Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
Source Data http://openspending.org open platform public finance data from governments around the world more than 350 datasets, more than 17 million transactions updated regularily (about a dataset a week) Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
Source Data "sub -programme": { "label":"Security and safeguarding liberties", "html_url":"http:// openspending.org/eu - budget/sub -programme/security -and - safeguarding -liberties", "name":"security -and -safeguarding -liberties " } , "html_url": "http:// openspending.org/eu - budget/entries/017dfcb58d05671ef9eb5a9f77 fef39c8b14150c", "amount": 41.2 Figure : simplified excerpt from an OpenSpending entry Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
Why Convert the Data to RDF? Problems with source data source data is structured (database) but not semantic open, but not linked data silo: own format, not interlinked to other knowledge bases, hard to integrate Benefits of RDF multiple ways of access for both machines and people: resolving of URIs, SPARQL, RDF dump use of Linked Open Vocabularies: common vocabulary → easier integration with other spending data semantic web infrastructure such as Question Answering Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
Problems with Conversions JSON API, no bulk download frequent changes errors in the data source data uses specific model for statistical observations: data cube big amount of data → performane is important to not have long waiting times common data model but datasets have different vocabulary Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
How the problems were solved Problems Solution JSON API custom Java program, defining JSON path expressions changes two step process: (1) download all JSON resources (2) convert JSON to RDF errors defining error rate thresholds for accept/decline of datasets data cube use of Linked Open Vocabulary RDF data cube model performance use of persistent caching different not yet solved, needs more time (cooperation vocabulary with experts, student thesis?) Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
Outcome: LinkedSpending Total Average number of datasets 247 filesize (RDF/N-Triples) 10 GB 41 MB triples 50 million 200 000 observations 2.4 million 10 000 Table : total and average values (approximate) available at http://linkedspending.aksw.org/ 1 RDF dump for access of the whole dataset public SPARQL endpoint for queries OntoWiki instance for browsing 1 still under development Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
information need SPARQL Query 1 all years which have observations in the s e l e c t d i s t i n c t ? date { ?o a qb : Observation . de-bund dataset from ?o qb : dataSet l s : de − bund . 2020 onwards ?o sdmxd : r e f P e r i o d ? date . FILTER ( xsd : date (? date ) > = ”2020 − 1 − 1”ˆˆxsd : date ) } 2 spendings of more than 100 billion e s e l e c t ∗ { ?o l s o : amount ?a . ?o dbo : c u r r e n c y dbpedia : Euro . FILTER ( xsd : i n t e g e r (? a) > ”1E11”ˆˆ xsd : i n t e g e r ) } 3 datasets with multiple years s e l e c t ?d count (? y ) as ? count { ?d a qb : DataSet . ?d l s o : r e f Y e a r ?y . } having ( count (? y) > 1) 4 sums of amounts for each reference year of s e l e c t ?y (sum( xsd : i n t e g e r (? amount ) ) as ?sum) the dataset berlin de { ?o qb : dataSet l s : b e r l i n d e . ?o l s o : r e f Y e a r ?y . ?o l s o : amount ?amou 5 datasets with curren- cies whose inflation s e l e c t d i s t i n c t ?d ? c ? r { ?o qb : dataSet ?d . ?o rate is greater than dbo : c u r r e n c y ? c . ? c dbp : i n f l a t i o n R a t e ? r . f i l t e r (? r > 10) } 10 % 6 Berlin city subsectors . . . of research and educa- tion that have had their budget reduced from 2012 to 2013 Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
Konrad H¨ offner konrad.hoeffner@uni-leipzig.de
Recommend
More recommend
